C++ WERD类代码示例

void make_single_word(bool one_blob, TO_ROW_LIST *rows, ROW_LIST* real_rows) {
  TO_ROW_IT to_row_it(rows);
  TO_ROW* row = to_row_it.data();
  // The blobs have to come out of the BLOBNBOX into the C_BLOB_LIST ready
  // to create the word.
  C_BLOB_LIST cblobs;
  C_BLOB_IT cblob_it(&cblobs);
  BLOBNBOX_IT box_it(row->blob_list());
  for (;!box_it.empty(); box_it.forward()) {
    BLOBNBOX* bblob= box_it.extract();
    if (bblob->joined_to_prev() || (one_blob && !cblob_it.empty())) {
      if (bblob->cblob() != NULL) {
        C_OUTLINE_IT cout_it(cblob_it.data()->out_list());
        cout_it.move_to_last();
        cout_it.add_list_after(bblob->cblob()->out_list());
        delete bblob->cblob();
      }
    } else {
      if (bblob->cblob() != NULL)
        cblob_it.add_after_then_move(bblob->cblob());
      delete bblob;
    }
  }
  // Convert the TO_ROW to a ROW.
  ROW* real_row = new ROW(row, static_cast<inT16>(row->kern_size),
                          static_cast<inT16>(row->space_size));
  WERD_IT word_it(real_row->word_list());
  WERD* word = new WERD(&cblobs, 0, NULL);
  word->set_flag(W_BOL, TRUE);
  word->set_flag(W_EOL, TRUE);
  word_it.add_after_then_move(word);
  ROW_IT row_it(real_rows);
  row_it.add_after_then_move(real_row);
}

WERD *add_repeated_word(                         //move repeated word
                        WERD_IT *rep_it,         //repeated words
                        int16_t &rep_left,         //left edge of word
                        int16_t &prev_chop_coord,  //previous word end
                        uint8_t &blanks,           //no of blanks
                        float pitch,             //char cell size
                        WERD_IT *word_it         //list of words
                       ) {
  WERD *word;                    //word to move
  int16_t new_blanks;              //extra blanks

  if (rep_left > prev_chop_coord) {
    new_blanks = (uint8_t) floor ((rep_left - prev_chop_coord) / pitch + 0.5);
    blanks += new_blanks;
  }
  word = rep_it->extract ();
  prev_chop_coord = word->bounding_box ().right ();
  word_it->add_after_then_move (word);
  word->set_blanks (blanks);
  rep_it->forward ();
  if (rep_it->empty ())
    rep_left = INT16_MAX;
  else
    rep_left = rep_it->data ()->bounding_box ().left ();
  blanks = 0;
  return word;
}

void PrintSegmentationStats(BLOCK_LIST* block_list) {
  int num_blocks = 0;
  int num_rows = 0;
  int num_words = 0;
  int num_blobs = 0;
  BLOCK_IT block_it(block_list);
  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
    BLOCK* block = block_it.data();
    ++num_blocks;
    ROW_IT row_it(block->row_list());
    for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
      ++num_rows;
      ROW* row = row_it.data();
      // Iterate over all werds in the row.
      WERD_IT werd_it(row->word_list());
      for (werd_it.mark_cycle_pt(); !werd_it.cycled_list(); werd_it.forward()) {
        WERD* werd = werd_it.data();
        ++num_words;
        num_blobs += werd->cblob_list()->length();
      }
    }
  }
  tprintf("Block list stats:\nBlocks = %d\nRows = %d\nWords = %d\nBlobs = %d\n",
          num_blocks, num_rows, num_words, num_blobs);
}

PAGE_RES_IT* make_pseudo_word(PAGE_RES* page_res, const TBOX& selection_box) {
  PAGE_RES_IT pr_it(page_res);
  C_BLOB_LIST new_blobs;               // list of gathered blobs
  C_BLOB_IT new_blob_it = &new_blobs;  // iterator

  for (WERD_RES* word_res = pr_it.word(); word_res != NULL;
       word_res = pr_it.forward()) {
    WERD* word = word_res->word;
    if (word->bounding_box().overlap(selection_box)) {
      C_BLOB_IT blob_it(word->cblob_list());
      for (blob_it.mark_cycle_pt();
           !blob_it.cycled_list(); blob_it.forward()) {
        C_BLOB* blob = blob_it.data();
        if (blob->bounding_box().overlap(selection_box)) {
          new_blob_it.add_after_then_move(C_BLOB::deep_copy(blob));
        }
      }
      if (!new_blobs.empty()) {
        WERD* pseudo_word = new WERD(&new_blobs, 1, NULL);
        word_res = pr_it.InsertSimpleCloneWord(*word_res, pseudo_word);
        PAGE_RES_IT* it = new PAGE_RES_IT(page_res);
        while (it->word() != word_res && it->word() != NULL) it->forward();
        ASSERT_HOST(it->word() == word_res);
        return it;
      }
    }
  }
  return NULL;
}

/// Builds a PAGE_RES from the block_list in the way required for ApplyBoxes:
/// All fuzzy spaces are removed, and all the words are maximally chopped.
PAGE_RES* Tesseract::SetupApplyBoxes(const GenericVector<TBOX>& boxes,
                                     BLOCK_LIST *block_list) {
  PreenXHeights(block_list);
  // Strip all fuzzy space markers to simplify the PAGE_RES.
  BLOCK_IT b_it(block_list);
  for (b_it.mark_cycle_pt(); !b_it.cycled_list(); b_it.forward()) {
    BLOCK* block = b_it.data();
    ROW_IT r_it(block->row_list());
    for (r_it.mark_cycle_pt(); !r_it.cycled_list(); r_it.forward ()) {
      ROW* row = r_it.data();
      WERD_IT w_it(row->word_list());
      for (w_it.mark_cycle_pt(); !w_it.cycled_list(); w_it.forward()) {
        WERD* word = w_it.data();
        if (word->cblob_list()->empty()) {
          delete w_it.extract();
        } else {
          word->set_flag(W_FUZZY_SP, false);
          word->set_flag(W_FUZZY_NON, false);
        }
      }
    }
  }
  PAGE_RES* page_res = new PAGE_RES(false, block_list, NULL);
  PAGE_RES_IT pr_it(page_res);
  WERD_RES* word_res;
  while ((word_res = pr_it.word()) != NULL) {
    MaximallyChopWord(boxes, pr_it.block()->block,
                      pr_it.row()->row, word_res);
    pr_it.forward();
  }
  return page_res;
}

// Remove outlines that are a tiny fraction in either width or height
// of the word height.
void Textord::clean_small_noise_from_words(ROW *row) {
  WERD_IT word_it(row->word_list());
  for (word_it.mark_cycle_pt(); !word_it.cycled_list(); word_it.forward()) {
    WERD* word = word_it.data();
    int min_size = static_cast<int>(
      textord_noise_hfract * word->bounding_box().height() + 0.5);
    C_BLOB_IT blob_it(word->cblob_list());
    for (blob_it.mark_cycle_pt(); !blob_it.cycled_list(); blob_it.forward()) {
      C_BLOB* blob = blob_it.data();
      C_OUTLINE_IT out_it(blob->out_list());
      for (out_it.mark_cycle_pt(); !out_it.cycled_list(); out_it.forward()) {
        C_OUTLINE* outline = out_it.data();
        outline->RemoveSmallRecursive(min_size, &out_it);
      }
      if (blob->out_list()->empty()) {
        delete blob_it.extract();
      }
    }
    if (word->cblob_list()->empty()) {
      if (!word_it.at_last()) {
        // The next word is no longer a fuzzy non space if it was before,
        // since the word before is about to be deleted.
        WERD* next_word = word_it.data_relative(1);
        if (next_word->flag(W_FUZZY_NON)) {
          next_word->set_flag(W_FUZZY_NON, false);
        }
      }
      delete word_it.extract();
    }
  }
}

//yangjing01 modified : 
bool TAL_make_single_word(bool one_blob, TO_ROW_LIST* rows, ROW_LIST* real_rows)
{
  TO_ROW_IT to_row_it(rows);
  ROW_IT row_it(real_rows);
  //to_real_row is the real row information of single row or single char mode
  TO_ROW* real_to_row = NULL;
  float row_max_height = 0.0;
  for (to_row_it.mark_cycle_pt();
    !to_row_it.cycled_list(); to_row_it.forward()){
    TO_ROW* row = to_row_it.data();
    float row_min_y = row->min_y();
    float row_max_y = row->max_y();
    float row_height = abs(row_max_y - row_min_y);
    if (real_to_row == NULL
      || row_height > row_max_height
      || fabs(row_height - row_max_height) < 1.0f){
      row_max_height = row_height;
      real_to_row = row;
    }
  }

  if (real_to_row == NULL){
    return false;
  }

  C_BLOB_LIST cblobs;
  C_BLOB_IT cblob_it(&cblobs);
  BLOBNBOX_IT box_it(real_to_row->blob_list());
  for (; !box_it.empty(); box_it.forward()){
    BLOBNBOX* bblob = box_it.extract();
    if (bblob->joined_to_prev() || (one_blob && !cblob_it.empty())) {
      if (bblob->cblob() != NULL){
        C_OUTLINE_IT cout_it(cblob_it.data()->out_list());
        cout_it.move_to_last();
        cout_it.add_list_after(bblob->cblob()->out_list());
        delete bblob->cblob();
      }
    }
    else {
      if (bblob->cblob() != NULL)
        cblob_it.add_after_then_move(bblob->cblob());
    }
    delete bblob;
  }
  // Convert the TO_ROW to a ROW.
  ROW* real_row = new ROW(real_to_row, static_cast<inT16>(real_to_row->kern_size),
    static_cast<inT16>(real_to_row->space_size));
  WERD_IT word_it(real_row->word_list());
  WERD* word = new WERD(&cblobs, 0, NULL);
  word->set_flag(W_BOL, TRUE);
  word->set_flag(W_EOL, TRUE);
  word->set_flag(W_DONT_CHOP, one_blob);
  word_it.add_after_then_move(word);
  row_it.add_after_then_move(real_row);

  return true;
}

WERD *make_real_word(BLOBNBOX_IT *box_it,  //iterator
                     inT32 blobcount,      //no of blobs to use
                     BOOL8 bol,            //start of line
                     uinT8 blanks          //no of blanks
                    ) {
  OUTLINE_IT out_it;             // outlines
  C_OUTLINE_IT cout_it;
  PBLOB_LIST blobs;              // blobs in word
  C_BLOB_LIST cblobs;
  PBLOB_IT blob_it = &blobs;     // iterator
  C_BLOB_IT cblob_it = &cblobs;
  WERD *word;                    // new word
  BLOBNBOX *bblob;               // current blob
  inT32 blobindex;               // in row

  for (blobindex = 0; blobindex < blobcount; blobindex++) {
    bblob = box_it->extract();
    if (bblob->joined_to_prev()) {
      if (bblob->blob() != NULL) {
        out_it.set_to_list(blob_it.data()->out_list());
        out_it.move_to_last();
        out_it.add_list_after(bblob->blob()->out_list());
        delete bblob->blob();
      }
      else if (bblob->cblob() != NULL) {
        cout_it.set_to_list(cblob_it.data()->out_list());
        cout_it.move_to_last();
        cout_it.add_list_after(bblob->cblob()->out_list());
        delete bblob->cblob();
      }
    }
    else {
      if (bblob->blob() != NULL)
        blob_it.add_after_then_move(bblob->blob());
      else if (bblob->cblob() != NULL)
        cblob_it.add_after_then_move(bblob->cblob());
    }
    delete bblob;
    box_it->forward();          // next one
  }

  if (blanks < 1)
    blanks = 1;

  if (blob_it.empty())
    word = new WERD(&cblobs, blanks, NULL);
  else
    word = new WERD(&blobs, blanks, NULL);

  if (bol)
    word->set_flag(W_BOL, TRUE);
  if (box_it->at_first())
    word->set_flag(W_EOL, TRUE);  // at end of line

  return word;
}

void ROW::plot(               //draw it
               ScrollView* window  //window to draw in
              ) {
  WERD *word;                    //current word
  WERD_IT it = &words;           //words of ROW

  for (it.mark_cycle_pt (); !it.cycled_list (); it.forward ()) {
    word = it.data ();
    word->plot (window);         //in rainbow colours
  }
}

/**
 * @name process_selected_words()
 *
 * Walk the current block list applying the specified word processor function
 * to each word that overlaps the selection_box.
 */
void Tesseract::process_selected_words(
    PAGE_RES* page_res, // blocks to check
    TBOX & selection_box,
    BOOL8(tesseract::Tesseract::*word_processor)(PAGE_RES_IT* pr_it)) {
  for (PAGE_RES_IT page_res_it(page_res); page_res_it.word() != NULL;
       page_res_it.forward()) {
    WERD* word = page_res_it.word()->word;
    if (word->bounding_box().overlap(selection_box)) {
      if (!(this->*word_processor)(&page_res_it))
        return;
    }
  }
}

void apply_box_training(BLOCK_LIST *block_list) {
  BLOCK_IT block_it(block_list);
  ROW_IT row_it;
  ROW *row;
  WERD_IT word_it;
  WERD *word;
  WERD *bln_word;
  WERD copy_outword;             // copy to denorm
  PBLOB_IT blob_it;
  DENORM denorm;
  INT16 count = 0;
  char ch[2];

  ch[1] = '\0';

  tprintf ("Generating training data\n");
  for (block_it.mark_cycle_pt ();
  !block_it.cycled_list (); block_it.forward ()) {
    row_it.set_to_list (block_it.data ()->row_list ());
    for (row_it.mark_cycle_pt (); !row_it.cycled_list (); row_it.forward ()) {
      row = row_it.data ();
      word_it.set_to_list (row->word_list ());
      for (word_it.mark_cycle_pt ();
      !word_it.cycled_list (); word_it.forward ()) {
        word = word_it.data ();
        if ((strlen (word->text ()) == 1) &&
        (word->gblob_list ()->length () == 1)) {
          /* Here is a word with a single char label and a single blob so train on it */
          bln_word =
            make_bln_copy (word, row, row->x_height (), &denorm);
          blob_it.set_to_list (bln_word->blob_list ());
          ch[0] = *word->text ();
          tess_training_tester (blob_it.data (),
                                 //single blob
            &denorm, TRUE,       //correct
            ch,                  //correct ASCII char
            1,                   //ASCII length
            NULL);
          copy_outword = *(bln_word);
          copy_outword.baseline_denormalise (&denorm);
          blob_it.set_to_list (copy_outword.blob_list ());
          ch[0] = *word->text ();
          delete bln_word;
          count++;
        }
      }
    }
  }
  tprintf ("Generated training data for %d blobs\n", count);
}

WERD *make_real_word(BLOBNBOX_IT *box_it,  //iterator
                     int32_t blobcount,      //no of blobs to use
                     bool bol,            //start of line
                     uint8_t blanks          //no of blanks
                    ) {
  C_OUTLINE_IT cout_it;
  C_BLOB_LIST cblobs;
  C_BLOB_IT cblob_it = &cblobs;
  WERD *word;                    // new word
  BLOBNBOX *bblob;               // current blob
  int32_t blobindex;               // in row

  for (blobindex = 0; blobindex < blobcount; blobindex++) {
    bblob = box_it->extract();
    if (bblob->joined_to_prev()) {
      if (bblob->cblob() != nullptr) {
        cout_it.set_to_list(cblob_it.data()->out_list());
        cout_it.move_to_last();
        cout_it.add_list_after(bblob->cblob()->out_list());
        delete bblob->cblob();
      }
    }
    else {
      if (bblob->cblob() != nullptr)
        cblob_it.add_after_then_move(bblob->cblob());
    }
    delete bblob;
    box_it->forward();          // next one
  }

  if (blanks < 1)
    blanks = 1;

  word = new WERD(&cblobs, blanks, nullptr);

  if (bol)
    word->set_flag(W_BOL, true);
  if (box_it->at_first())
    word->set_flag(W_EOL, true);  // at end of line

  return word;
}

void RefreshWordBlobsFromNewBlobs(BLOCK_LIST* block_list,
                                  C_BLOB_LIST* new_blobs,
                                  C_BLOB_LIST* not_found_blobs) {
  // Now iterate over all the blobs in the segmentation_block_list_, and just
  // replace the corresponding c-blobs inside the werds.
  BLOCK_IT block_it(block_list);
  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
    BLOCK* block = block_it.data();
    if (block->poly_block() != NULL && !block->poly_block()->IsText())
      continue;  // Don't touch non-text blocks.
    // Iterate over all rows in the block.
    ROW_IT row_it(block->row_list());
    for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
      ROW* row = row_it.data();
      // Iterate over all werds in the row.
      WERD_IT werd_it(row->word_list());
      WERD_LIST new_words;
      WERD_IT new_words_it(&new_words);
      for (werd_it.mark_cycle_pt(); !werd_it.cycled_list(); werd_it.forward()) {
        WERD* werd = werd_it.extract();
        WERD* new_werd = werd->ConstructWerdWithNewBlobs(new_blobs,
                                                         not_found_blobs);
        if (new_werd) {
          // Insert this new werd into the actual row's werd-list. Remove the
          // existing one.
          new_words_it.add_after_then_move(new_werd);
          delete werd;
        } else {
          // Reinsert the older word back, for lack of better options.
          // This is critical since dropping the words messes up segmentation:
          // eg. 1st word in the row might otherwise have W_FUZZY_NON turned on.
          new_words_it.add_after_then_move(werd);
        }
      }
      // Get rid of the old word list & replace it with the new one.
      row->word_list()->clear();
      werd_it.move_to_first();
      werd_it.add_list_after(&new_words);
    }
  }
}

/**
 * Returns the baseline of the current object at the given level.
 * The baseline is the line that passes through (x1, y1) and (x2, y2).
 * WARNING: with vertical text, baselines may be vertical!
 */
bool PageIterator::Baseline(PageIteratorLevel level,
                            int* x1, int* y1, int* x2, int* y2) const {
  if (it_->word() == NULL) return false;  // Already at the end!
  ROW* row = it_->row()->row;
  WERD* word = it_->word()->word;
  TBOX box = (level == RIL_WORD || level == RIL_SYMBOL)
           ? word->bounding_box()
           : row->bounding_box();
  int left = box.left();
  ICOORD startpt(left, static_cast<inT16>(row->base_line(left) + 0.5));
  int right = box.right();
  ICOORD endpt(right, static_cast<inT16>(row->base_line(right) + 0.5));
  // Rotate to image coordinates and convert to global image coords.
  startpt.rotate(it_->block()->block->re_rotation());
  endpt.rotate(it_->block()->block->re_rotation());
  *x1 = startpt.x() / scale_ + rect_left_;
  *y1 = (rect_height_ - startpt.y()) / scale_ + rect_top_;
  *x2 = endpt.x() / scale_ + rect_left_;
  *y2 = (rect_height_ - endpt.y()) / scale_ + rect_top_;
  return true;
}

void ExtractBlobsFromSegmentation(BLOCK_LIST* blocks,
                                  C_BLOB_LIST* output_blob_list) {
  C_BLOB_IT return_list_it(output_blob_list);
  BLOCK_IT block_it(blocks);
  for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {
    BLOCK* block = block_it.data();
    ROW_IT row_it(block->row_list());
    for (row_it.mark_cycle_pt(); !row_it.cycled_list(); row_it.forward()) {
      ROW* row = row_it.data();
      // Iterate over all werds in the row.
      WERD_IT werd_it(row->word_list());
      for (werd_it.mark_cycle_pt(); !werd_it.cycled_list(); werd_it.forward()) {
        WERD* werd = werd_it.data();
        return_list_it.move_to_last();
        return_list_it.add_list_after(werd->cblob_list());
        return_list_it.move_to_last();
        return_list_it.add_list_after(werd->rej_cblob_list());
      }
    }
  }
}